A Terrain Referenced Navigation Based on LiDAR Breakline Matching

J. Oh, C.K. Toth, D.A. Grejner-Brzezinska

Abstract:	Navigation based on GPS/INS integration requires a backup system to support robust navigation in the GPSdenied environment. Terrain referenced navigation, referred to as TRN, is one of the approaches that has recently gained more interest, due to more widespread availability of sensors that can directly sense terrain while the platform carrying the sensor is in motion. In TRN, the sensed terrain information is correlated with the existing reference data to derive the navigation estimate. With the advance in the geospatial sensor technology, highperformance ranging device, LiDAR (light detection and ranging), was developed showing its potential for TRN due to its high accuracy, high resolution, and high update rate. In addition, LiDAR can acquire not only terrain data, but also intensity response. Note that high terrain undulation should provide good constraint for accurate TRN while intensity information can be useful over the smooth terrain, where the LiDAR point cloud may not provide for reliable navigation due to the lack of terrain features. This study proposes the use of complementary terrain and intensity information in LiDAR data for robust and accurate TRN of airborne platforms. When utilizing intensity information, one concern is the significant spectral difference between the sensed LiDAR data and the reference data due to the difference in data acquisition sensors, including date, time, and weather. In the approach proposed here, this difficulty is overcome utilizing a robust breakline matching, called Relative Edge Cross Correlation (RECC). Since RECC is a ‘relative’ indicator for image matching results in contrast to the conventional normalized cross correlation matching which uses an absolute value as a threshold, the concentration value (CV) is introduced to determine the correct matching candidate among multiple matching candidates. In the proposed method, the target situation is that LiDAR data are acquired under GPS-denied environment and reference data of the area are available onboard. First, both intensity and terrain height information of sensed LiDAR data and reference data are converted to an image (from point cloud to raster format), followed by breakline extraction. Then, RECC image matching is performed between intensity images and between terrain height images, followed by CV computation to obtain reliable matching points. Utilizing the best result among the intensity and terrain height matching, three-dimensional positional fixes can be obtained. This paper describes the concept and the procedures of this TRN approach together with the experimental results. The test LiDAR data were acquired over San Diego at about 2,000 m altitude (ASL), with a point density ranging between 0.6 and 3 pts/m2. Two different reference data sets were tested: LiDAR data and high-resolution satellite imagery (HRSI). The experimental results showed mean positional accuracy better than 50 cm when LiDAR data is used as reference, and approximately 1.6 m for the HRSI reference.
Published in:	Proceedings of the 2011 International Technical Meeting of The Institute of Navigation January 24 - 26, 2011 Catamaran Resort Hotel San Diego, CA
Pages:	868 - 879
Cite this article:	Oh, J., Toth, C.K., Grejner-Brzezinska, D.A., "A Terrain Referenced Navigation Based on LiDAR Breakline Matching," Proceedings of the 2011 International Technical Meeting of The Institute of Navigation, San Diego, CA, January 2011, pp. 868-879.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In